Overview of “Update of AAPM Task Group
No. 43 Report - A Revised AAPM Protocol
for Interstitial Brachytherapy Dose
Calculations”
Jeffrey F. Williamson, Ph.D.
On behalf of Report Authors
Mark Rivard, Bert Coursey, Larry DeWerd, William
Hanson Saiful Huq, Geoffrey Ibbott, Michael Mitch,
Ravinder Nath, Jeff Williamson
VCU Radiation Oncology Physics
Virginia Commonwealth University
Outline
• Motivation of a revised TG-43
• Report organization & goals
• Revised TG-43 Content review
– Calibration standards and dose estimation technology: review and report
recommendations
– Modified TG-43 formalism
– Data merging and analysis strategy
– Clinical implementation issues
• Learning Objectives
– Understand basic concepts and issues in low-energy source dosimetry
– Understand TG-43 formalism and important recommendations; and major
differences between 1995 and 2003 editions
Single-Source Dose Distributions
Superposition Model
θ
r
Isodoses
100 cGy/h
50
20
10
7
5
3
1
0.5
Current dose calculation paradigm: for each source type, a
single source dose array used to estimate the contribution of
each source, based on coordinates in patient, to each point of
interest
Basic Elements of TG-43 DoseCalculation ‘Algorithm’
• For each source, a discrete grid of measured or
Monte Carlo dose rates is required
• A set of standard dose ratios (Λ, F(r,θ), g(r ), ϕan(r ) )
are defined to represent measured dose-rate
distribution in compact form
• A standard approach for interpolating between table
entries to get dose rate at an arbitrary point (r,θ)
Starting Point: discrete grid of dose rates
measured by TLD or calculated by Monte Carlo
Transverse Axis Measurement Phantom
Polar Dose Profile Measurement Phantom
90
120
135
160
180
o
o
o
60
o
45
o
o
o
20
o
0
o
Why do we need a new TG 43 report?
• Report originated and written by AAPM Low-energy
Interstitial Brachytherapy Dosimetry SC (LIBD) of RTC
– The number of seed models has increased from 3 in 1999 to
approximately 20 in 2003.
» 1995 report limited to 6711, 6702 and Model 200 Pd.
» Data is old and needs to be supplanted by new
– Correct errors/inconsistencies in old formalism and
definitions
– Present up-to-date consensus dosimetry sets for 8 seed
models
– Provide guidance and standards for MC and TLD
investigators
– Consolidate recent LIBD guidance in single report
Revised TG-43 Contents
• III: updated spectral and half-life data for 125I
&103Pd
• IV: History of 1985 (Loftus) and 1999 (WAFAC)
calibration standards
– Revised SK definition
• V: Revised 2-D and 1-D dose calculation
formalisms
• VI: Consensus data sets for clinical
implementation
– Procedure for comparing and merging multiple TLD and
MC data sets from literature
– Tables of TG-43 parameters
Revised TG-43 Contents: cont’d
• VI: Consensus data sets for clinical implementation
– Sources covered by this report (LIBD compliant as of 7/01)
» Amersham-Health model 6702 and 6711 125I sources
» Best Industries model 2301 125I source
» North American Scientific Inc. model MED3631-A/M 125I
source
» Bebig model I25.SO6 125I source
» Imagyn Medical Technologies Inc. isostar model IS12501 125I source
» Theragenics model 200 103Pd source
» North American Scientific Inc. model MED3633 103Pd
source
Revised TG-43 Contents: cont’d
• VII: Recommendations for estimation and tabulation
of dosimetry parameters
– Minimum standards: angular/distance range and
resolution
– Guidelines for TLD dosimetry
– Guidelines for Monte Carlo dosimetry
• VIII: Clinical implementation
– Acceptance testing and commissioning
– Guidelines for calibration and traceability
• Appendices
– Extrapolation to large and small distances
– Use of apparent activity
– Anisotropy constant
AAPM Revised 2-D Formalism
θ1°
r
1.687
D(r, θ) = SK ⋅ Λ ⋅
G L (r, θ)
⋅ F(r, θ) ⋅ g L (r)
G L (1 cm, π / 2)
 β
if θ ≠ 0°
 Lr sin θ
for x = L G L (r, θ) = 
 r 2 − L2 / 4 −1 if θ = 0°

(
for x = P
)
G P (r, θ) = r −2
Where L = effective active length of source
line-source
approximation
point-source
approximation
Task Group 43 2-D Radial Dose Function: gX(r)
1.20
1.00
Model 6711: Monte Carlo
Radial Dose Function: g(r)
• gX(r) = dimensionless radial
dose function Describes
transverse-axis dose Fall-off
• GX-Factor suppresses dose
variation due to inverse squarelaw fall-off
D(r, π /2) ⋅ G X (1 cm, π /2)
g X (r) =
D(1 cm, π /2) ⋅ G X (r, π /2)
g P (r) =
D(r, π /2)  r 
⋅

D(1 cm, π /2)  1 cm 
2
Model 6702: Monte Carlo
0.80
Model 6711: TLD
Model 6702 : TLD
0.60
0.40
0.20
0.00
0.0
2.0
4.0
6.0
8.0
Distance (cm)
Task Group 43 2-D Anisotropy Function: F(r,θ)
• F(r,θ) = dimensionless
anisotropy function
– Describes angular dose variation
at fixed distance
– G suppresses inverse-square
dose variation
– Only Line-source G recommended
F(r, θ) =
D(r, θ) ⋅ G L (r, π /2)
D(r, π /2) ⋅ G L (r, θ)
θ1°
r
1.687
10.0
2.60
2.60
DraxImage Model LS-1
125
2.40
I Source
Double Point Source G(r,θ)
2.20
DraxImage Model LS-1
2.00
125
I Source
Single Point Source G(r,θ)
2.20
2.00
Polar Dose profile
Anisotropy Function: F(r,θ)
2.40
1.80
1.60
1.40
0.25 cm
1.5 cm
0.5 cm
2 cm
0.75 cm
5 cm
1 cm
10 cm
1.20
1.60
1.5 cm
2.0 cm
5.0 cm
10.0 cm
1.40
1.20
1.00
1.00
0.80
0.80
0.60
0.25 cm
0.5 cm
0.75 cm
1.0 cm
1.80
0.60
0
15
30
45
60
75
90
0
Polar Angle
15
30
45
60
75
Polar Angle
Revised TG 43 1-D Formalism
D(r) = SK ⋅ Λ ⋅
θ
G X (r, π/2)
⋅ g (r) ⋅ φan (r)
G X (1 cm, π /2) X
53.1°
•
φan(r) is 1-D anisotropy function
r
1.687
φan (r) =
π
D(r, θ) ⋅ sin θ ⋅ dθ
0
∫
2 ⋅ D(r, π / 2)
90
TG 43 Update: Two possible 1D Models
• End user’s RTP system MUST use the
Geometry function used to extract the TG-43
quantities from raw data
– P = Point Source (Acceptable option)
2
1 cm 
D(r) = SK ⋅ Λ ⋅ 
 ⋅ g P (r) ⋅ φan (r)
 r 
– L = Line Source (Recommended option)
D(r) = SK ⋅ Λ ⋅
G L (r, π/2)
⋅ g (r) ⋅ φan (r)
G L (1 cm, π /2) L
TG-43 Dose-Calculation ‘Algorithm’
• TG-43 starts with a discrete grid of measured or
Monte Carlo dose rates
• F(r,θ), g(r ), and ϕan(r ) table entries correspond
to measurement/MC points
• What does RTP do at arbitrary point (r,θ)?:
– Finds g(r1) and g(r2), etc. at nearest neighbor points
– Calculates g(r ), etc., by bi-linear interpolation
 r − r1 
g(r) = [ g(r2 ) − g(r1 )] 
 + g(r1 )
 r2 − r1 
– Calculate exact G(r,θ) and obtain D(r,θ) from TG-43
equation
TG 43 Updated Formalism
• Extrapolation to large and small distances outside
table entries addressed
• Anisotropy constant discarded
Trick RTP system by using: g'(r) ⇒ g X (r) ⋅ φan (r)
• Rationale: Geometry function recommendations
– Any G, if consistently applied, will always reproduce
exactly the measured data
– More accurate G ⇒ more accurate interpolation between
table entries
– GL is a good compromise between simplicity and accuracy
– More complex G acceptable but not included in report
Revised Definition of Air-Kerma Strength
SK = K δ (d) d 2 [µGy ⋅ m 2 ⋅ h −1 = cGy ⋅ cm 2 ⋅ h −1 = U]
K δ (d) is air-kerma rate in vacuo due to
photons of energy > δ ( ∼ 5 keV), d
L
Cutoff designed to exclude low-energy contaminant
radiation
Ritz Free-Air Chamber
NIST 1985 SK Standard for 125I Seeds
• Implemented in 1985 for 6711 and 6702 seeds;
difficult to extend to new seed models
• No filtering of Ti K x rays (4.5 keV)
• Small volume ⇒ only seed plaques measurable
Influence of Ti K-edge X-rays on SK measurements
NIST 1985 SK Standard for 125I Seeds
J. Williamson, Monte Carlo Calculations 1988
‘WAFAC:’ Wide Angle Free-Air Chamber
U.S. NIST Primary Standard for Low-Energy Seeds
Rotating Seed
Holder
Wide-angle Free-Air Chamber SK,N99
250 mm diameter
153mm Long
for same seed and
radiation output
Collecting volume
300mm
150mm
100mm
80mm
Source
0.08 mm Al filter
V
1 mm thick
tungsten collimator
V/2
S K,N99
S K,N85
0 (Guard
Ring)
I
•
•
•
•
Large volume ⇒ single seeds measurable
Extended to all 125I and 103Pd models
Ti characteristic x-rays filtered out
Implemented 1999
= 0.897
TG-43 Formalism: original = revised Λ
D(r, θ) = SK ⋅ Λ ⋅
G L (r, θ)
⋅ F(r, θ) ⋅ g L (r)
G L (1 cm, π / 2)
• SK is Air-Kerma Strength
θ1°
1 U = 1 Unit of Air-kerma strength
= 1 µGy•m2•h-1 = 1 cGy•cm2•h-1
• Λ =Dose-Rate Constant in
Λ=
r
1.687
cGy•h-1•U-1
D(1 cm, π / 2)
SK
Solid “water substitute” Phantoms for TLD Dosimetry
θ
53.1°
Transverse Axis Measurement Phantom
r
1.687
Λ=
Dwat at
{
r = 1 cm
θ =π/2
SK,N99
where Dwat = TLD measurement
SK,N99 = NIST measurement
Basic Discrete Event Monte Carlo Algorithm
Randomly select
Location, direction
& energy of primary
photon
Photon
Collision
select distance to
next collision
Select type of
collision
Select type of
collision
Heterogeneity
Ag core
Ti capsule
Scoring
Bin, ∆V
I-125 Seed
Select Energy and
angle of photon
leaving collision
Calculation of TG-43 Parameters
by MCPT
MCPT calculates per disintegration within source:
- Dose to medium, ∆Dmed(r), near source in phantom
geometry
- Air-kerma strength, ∆SK, in free-air geometry
Λ=
∆Dwat (r = 1 cm, θ = π / 2)
∆SK
g(r) =
∆Dwat (r, π / 2) ⋅ G(1 cm, π / 2)
∆Dwat (1 cm, π / 2) ⋅ G(r, π / 2)
WAFAC Simulation Method
( ∆E153
− ∆E11
) ⋅ d2
ab
ab
∆S K =
⋅ k inv ⋅ k att
ρair ⋅ (V153 − V11 )
x
= Energy absorbed/disintegration in WAFAC volume of length x
where ∆Eab
d = 38 cm = seed-to-WAFAC volume center


1.032 Pd-103
k att =
 for a point source = 1.023 I-125
2
k inv ⋅ ∆K ⋅ d

WFC 
{
( ∆SK )extr
(
)
k inv = inverse−square correction =
∫ Φ(
A
) ⋅ dA
Φ (d) ⋅ A
= 1.0089
General Recommendations
• Measure/calculate dose-rates with sufficient
resolution and range
– RDF: 0.5 cm to 5 cm (Pd-103) – 7 cm (I-125)
– F(r,θ): for r =1, 2, 3, 5 and 7 cm (I-125 only) at 10° intervals
– Report active length, L, used
– Perform rigorous uncertainty analysis
• Appropriate tools
– TLD-100/diode accepted for relative, TLD for absolute
» Other detectors: require extensive & published
benchmarking
– Well characterized MC code: EGS-4, MCNP, PTRAN
» New users/features/codes: reproduce published
benchmarks
Recommendations for TLD dosimetry
• For DRC: ≈ 15 measurements on 8-10 seeds, 1
with directly traceable NIST calibration
– 1-σ statistics < 5%, 1-σ systematic < 7%
– Calibrate TLD’s against external beam, not a
brachytherapy source
– Make sure E(d) matches experiment wrt calibration
technique, displacement/attenuation corrections,
measurement medium
– Correct for non-water equivalence of phantom: Solid
Water may require chemical analysis
– Recommended standards of methodology description
for published papers
Recommendations for Monte Carlo Dosimetry
• Monte Carlo Code recommendations
– Physics model: electron binding corrections, coherent
scattering and characteristic x-ray production. Electron
transport unnecessary
– NIST PHOTEX or equivalent cross-section library
– 1-s statistical uncertainty <2%, volume-averaging <1%
– Calculate dose rates in 30 cm diameter liquid water sphere
• Other precautions
– transverse-plane anisotropy ⇒ Simulate WAFAC aperture
– Carefully validate geometric model through mechanical
and radiological measurements. Perform parametric
studies to assess component motion, uncertainties, etc.
6711 silver rod end
Electron microscopy
Geometric Model
Validation
DraxImage I-125 Seed
6711 contact radiographs
Contact
Radiograph
Final Model
Data Merging Procedures: Λ
• 2-25 papers available documenting dosimetry for
each source model. Candidate datasets identified
– MC: NIST PHOTEX equivalent cross sections,
normalization to SK,N99 standard, and simulation of
WAFAC geometry for “sharp edges” phenomenon
– TLD: prefer recent studies based on SK,N99.
– All data: corrected to SK,N99 standard, for CY-2000 WAFAC
measurement errors, solid-to-liquid water corrections
– All data corrected to consistent active length, L
Λ con = 1  Λ exp + Λ MC 
2
where Λ X = mean of candidate exp or MC values
Data Merging: g(r), F(r,θ) & ϕan(r)
• Transform different candidate datasets to same L
• Graphically compare different datasets
• If discrepancies < experimental error, accept as
consensus, dataset with maximum range,
resolution and smoothness
– For g(r) and F(r,θ), usually MC data accepted
– For g(r), discrepancies < 5% for r< 5 cm
– F(r,θ), discrepancies < 10%, & < 5% for θ > 30º
• ϕan(r): calculate dose from consensus F(r,θ) and
integrate wrt to dΩ
• If discrepancies > experimental error, more
elaborate merging strategies needed
Example: Theragenics Model 200
Pd-103 Source
– <1994 = Heavy Seed: reactor-produced 20 µm
Pd layer on carbon pellet
– >1994 = Light Seed: accelerator-produced 2
µm layer
• No SK standard until 1999: Vendor Aapp
“standard” 1988-1999
0.560
3.140
4.500
0.890
• All Pd-103 products depend on Mod 200
History for dose prescription
• Two variants : “Heavy” and “Light” seeds
0.612
1.090
• Sole product 1987-1999
0.826
0.510
Pb marker
Graphite pellet
Model 200: Consensus Λ
• Candidate datasets
– MC: Monroe & Williamson: Med. Phys., 29:609-621, 2002
– TLD: Nath et al. Med Phys 27 (4), 655-658, 2000
1
x = 99 


Λ con =  Λ Nath,TLD ⋅  S K,N99 :
 + Λ JFW,MC 
2
x = 00 


1
+ 0.691
= [
0.65 ⋅
1.053
]
2
= 0.687
Model 200 Pd-103: radial dose function
1.5
1.20
MC/Meigonni
1.10
0.9
1.00
0.6
0.90
0.3
0.80
g(r) Meigooni 1990
g(r) Chiu-Tsao 1991
g(r) Monroe-Williamson 2002
0.0
0.70
0
1
2
3
Distance (cm)
4
5
6
Monte Carlo/Measured
Radial Dose Function
MC/Chiu-Tsao
1.2
• No “light seed” TLD
• Compare Meigooni 90
& Chiu-Tsao 91 TLD to
Monroe 02 Monte Carlo
• Monte Carlo: little
difference between
light and heavy seeds
• Monroe g(r) is
consensus dataset
Model 200: MC vs TLD anisotropy at 2 cm
TLD vs Monte Carlo Anisotropy at 2 cm
1.1
1.15
1.10
0.8
1.05
0.7
1.00
0.6
0.95
0.5
Monte Carlo/TLD
0.9
Anisotropy Function
• Agreement
within 5%
1.20
Yue-Nath 2002 TLD
Monroe-Williamson MC 2002
1
0.90
MC/TLD
0.4
0.85
0.3
0
15
30
45
60
75
90
0.80
Polar Angle
• ϕan(1 cm) =
0.859 (TLD)
0.855 (MC) vs
0.90 TG-43
• Use Monroe
TLD data (from
0.25 to 7 cm)
as consensus
data
Revised Dose-Rate Constants
Source
1995
TG-43
Λ95D,N85S
SK,N85
2000 LIBD
Guidance
Λ95D,N99S
SK,N99
Revised
TG-43
Λ03D,N99S
SK,N99
Model 6711
I-125
0.88
0.981
0.964
Model 6702
I-125
0.93
1.037
1.036
0.93
1.037
1.02
0.665
0.686
NAS 3631 A/S
I-125
Model 200
Pd-103
(Wallace 1998)
0.74
SK,T88
Conclusions and Future Work
• A major TG-43 revision is approaching approval
– Revised dosimetry parameters for 8 seed models
– Updated dose calculation formalism
– Guidance for future reference-quality dosimetry
• Future activities of LIBD
– Consensus data for remaining low energy sources
– Revised prescribed-to-administered dose ratios for Pd-103
– Consensus data for HDR and LDR Ir-192 sources
– Dose-calculation guidance for Cs-137 sources
– Procedures for vendor and end-user calibration procedures